sampling and investigating hard datatiiciiitm.com/profanurag/sampling.pdf · sampling data on...

Sampling and

Investigating

Hard Data

2

Major Topics

Sampling

Hard data

Qualitative document analysis

Workflow analysis

Business process reengineering

Archival documents

3

What is Sampling?

Sampling is a process of systematically

selecting representative elements of a

population

Key issues?

Which of the key documents and Web sites

should be sampled?

Which people should be interviewed or sent

questionnaires?

4

Why do we need Sampling?

The reasons systems analysts do sampling are

Reduction of costs

(copying documents and talking to everyone)

Speeding up the data-gathering process

(Due to selective data processing)

Improving effectiveness

(Talking to few people but asking detailed questions)

Reduction of data-gathering bias

(views of the executives, who have already handled the

existing IS, may provide biased evaluation)

5

Sampling Design Steps

To design a good sample, a systems analyst needs to follow four steps: Determining the data to be collected or described (only relevant data and

the methods of analysis))

Determining the population to be sampled ( Whom to interview? Not only the

employee but also the customer sometimes)

Choosing the type of sample (1&2 Unrestricted; 3&4 Selective)

1-Cnvenience or unrestricted, e.g. System Analysts may call a meeting on any issue

(Reliability ?);

2-Simple Random - equal chance to the each people to be questioned (non practical)

3- Purposive- based on judgment after meeting with few knowledgeable people

(Reliability?)

4-Complex random- (Systematic, Stratified (making subgroups and taking samples) and

clustered (group of documents or people to study))

6

Deciding the Sample Size

Sometimes by knowing the population thinking in a single way and having certain characteristics and sometimes by the mistakes in the forms- referred as attribute data

The sample size decision should be made according to the specific conditions under which a systems analyst works with such as

Sampling data on attributes

(Subjective decisions e.g. Confidence level)

Sampling data on variables

(On actual numbers e.g. gross sales, items

returned, number of mistakes)

Sampling qualitative data

( by interviewing and not through searching in files,

reports or documents)

7

Types of Sampling

There are four types of sampling

Convenience

Purposive

Simple random

Complex random

8

Convenience Sampling

Convenience samples are unrestricted,

nonprobability samples

Easy to arrange

Most unreliable

9

Simple Random Sampling

Based on a numbered list of the population

Each person or document has an equal

chance of being selected

Hence Non Practical

10

Purposive Sampling

Based on judgment

Analyst chooses group of individuals to

sample

Based on criteria

Nonprobability sample

Moderately reliable

11

Complex Random Sampling

Has three forms

Systematic sampling

Stratified sampling

Cluster sampling

12

Feasibility and Managing Analysis and Design Activities

Systematic Sampling

Simplest method of probability sampling

Choose every kth person on a list

Not good if the list is ordered

13


Stratified Sampling

Identifying subpopulations or strata

Selecting objects or people for sampling from

the subpopulation

Compensates for a disproportionate number

of employees from a certain group

Most important to the systems analyst

14


Cluster Sampling

Select group of documents or people to study

Select typical groups that represent the

remaining ones

15


Deciding Sample Size for

Attribute Data

Seven steps to determine sample size

Determine the attribute to sample

Locate the database or reports where the

attribute is found

Examine the attribute and estimate p, the

proportion of the population having the attribute

16



Attribute Data

Steps to determine sample size (continued)

Make the subjective decision regarding the

acceptable interval estimate, i

Choose the confidence level and look up the

confidence coefficient (z value) in a table

17


iσp =

z


Attribute Data


Calculate σp, the standard error of the proportion

as follows:

18


p(1-p)n = + 1

σp2


Attribute Data


Determine the necessary sample size, n, using

the following formula:

19


Confidence Level Table

99% 2.58

98% 2.33

97% 2.17

96% 2.05

95% 1.96

90% 1.65

80% 1.28

50% .67

20


Sample Size for Data on

Variables

The steps to determine the sample size when

sampling data on variables are

Determine the variable you will be sampling

Locate the database or reports where the variable

can be found

21



Variables

The steps to determine variable sample size

(continued)

Examine the variable to gain some idea about its

magnitude and dispersion

It would be useful to know the mean to determine a

more appropriate interval estimate i and the standard

deviation, s to determine sample size (in the last

step)

22



Variables


(continued)

Make a subjective decision regarding the

acceptable interval estimate, i

Choose a confidence level and look up the

confidence coefficient (z value)

23



Variables


(continued)

Calculate σx, the standard error of the mean as

follows:

iσx =

z

24



Variables


(continued)

Determine the necessary sample size, n, using

the following formula:

sn = + 1

σx2

2

25


Hard Data

In addition to sampling, investigation of hard

data is another effective method for systems

analysts to gather information

26


Obtaining Hard Data

Hard data can be obtained by

Analyzing quantitative documents such as

records used for decision making

Performance reports

Records

Data capture forms

Ecommerce and other transactions

27


Qualitative Documents

Examine qualitative documents (includes memos,

signs, bulletin boards, manual, policy handbook etc.) for the

following:

Key or guiding metaphors

Insiders vs. outsiders mentality

What is considered good vs. evil

Graphics, logos, and icons in common areas or

Web pages

A sense of humor

28


Analyzing Qualitative

Documents

Qualitative documents include

Memos

Signs on bulletin boards

Corporate Web sites

Manuals

Policy handbooks

29


Workflow Analysis

Workflow analysis may reveal signs of larger

problems, such as Data or information doesn’t flow as intended

Bottlenecks in the processing of forms

Access to online forms is cumbersome

Unnecessary duplication of work occurs reason being

employees are unaware that information is already in

existence

Employees lack understanding about the

interrelatedness of information flow

30


Business Process

Reengineering

Business process reengineering software

includes the following features:

Modeling of the existing system

Analysis of possible outcomes

Simulation of proposed work flow

31


Archival Documents

A systems analyst may obtain some valuable

information by abstracting data from archival

documents

Generally, archival documents are historical

data, and they are prepared and kept by

someone else for specific purposes

32


Guidelines for Abstracting

Archival Data

Fragment data into subclasses and make

cross-checks to reduce errors

Compare reports on the same phenomenon

by different analysts

Realize the inherent bias associated with

original decisions to file, keep, or destroy

reports

Use other methods to obtain data